Executive ‘‘Brake Failure’’ following Deactivationof Human Frontal Lobe
Christopher D. Chambers1, Mark A. Bellgrove1,2, Mark G. Stokes1,Tracy R. Henderson1, Hugh Garavan2, Ian H. Robertson2,
Adam P. Morris1, and Jason B. Mattingley1
Abstract
& In the course of daily living, humans frequently encountersituations in which a motor activity, once initiated, becomesunnecessary or inappropriate. Under such circumstances, theability to inhibit motor responses can be of vital importance.Although the nature of response inhibition has been studied inpsychology for several decades, its neural basis remainsunclear. Using transcranial magnetic stimulation, we found
that temporary deactivation of the pars opercularis in the rightinferior frontal gyrus selectively impairs the ability to stop aninitiated action. Critically, deactivation of the same region didnot affect the ability to execute responses, nor did it influencephysiological arousal. These findings confirm and extendrecent reports that the inferior frontal gyrus is vital formediating response inhibition. &
INTRODUCTION
The ability to apply executive control over actions isessential for normal human activities. Executive func-tions enable us to plan, execute, and update behaviorin response to an environment of continual change(Heyder, Suchan, & Daum, 2004; Logan, 1994). Inparticular, unexpected events frequently require us tocancel intended actions. Without the ability to inhibitand update motor activities, many aspects of everydayliving would become impossible, such as driving avehicle, undertaking sporting activities, and engagingin social interactions. The importance of motoric inhibi-tion as a core executive function is highlighted by thebroad range of psychiatric conditions that are character-ized by inhibitory deficits; among others, these includeobsessive–compulsive disorder (Enright & Beech, 1993),attention deficit hyperactivity disorder (ADHD) (Aron,Dowson, Sahakian, & Robbins, 2003), and schizophrenia(Badcock, Michie, Johnson, & Combrinck, 2002).
Although the cognitive mechanisms underlying re-sponse inhibition have been studied in experimentalpsychology for many years (Logan, 1981, 1994), keyquestions remain concerning its underlying neuralmechanisms. Most cognitive neuroscientists agree thatthe human prefrontal cortex is responsible for executivecontrol, but it is contentious whether discrete prefrontalregions are specialized to carry out domain-specificfunctions (Aron, Robbins, & Poldrack, 2004; Duncan
& Owen, 2000; Rowe, Toni, Josephs, Frackowiak, &Passingham, 2000; Goldman-Rakic, 1987). Some studieshave suggested that different prefrontal regions sharecontrol over a range of cognitive processes, includingthose involved in the inhibition and selection of re-sponses (Duncan & Owen, 2000). Others, however, haveargued that mechanisms of response inhibition aregoverned by a discrete network of brain regions inthe parietal and prefrontal cortex (Morita, Nakahara, &Hayashi, 2004; Aron, Fletcher, Bullmore, Sahakian,& Robbins, 2003; Rubia, Smith, Brammer, & Taylor,2003; Garavan, Ross, & Stein, 1999). Neurophysiologicalstudies in macaques, for instance, have revealed contri-butions of ventral prefrontal cortex to the suppressionof manual and saccadic responses (Hasegawa, Peterson,& Goldberg, 2004; Sakagami et al., 2001). In humans,neuroimaging studies have revealed selective activationof the inferior frontal gyrus (IFG), middle frontal gyrus(MFG), and inferior parietal cortex of the right hemi-sphere during inhibition of an intended action (Rubiaet al., 2003; Swainson et al., 2003; Garavan et al., 1999;Konishi et al., 1999; Kawashima et al., 1996). Further-more, a recent neuropsychological study showed thatlesions of the right IFG were predictive of inhibitorydeficits in patients with brain damage (Aron, Fletcher,et al., 2003).
Despite making a vital contribution to the cognitiveneuroscience of response inhibition, previous neuroim-aging and neuropsychological studies nevertheless havefundamental limitations. In particular, neuroimagingtechniques cannot distinguish between neural activity1University of Melbourne, 2Trinity College Dublin
D 2006 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 18:3, pp. 444–455
that is necessary for a behavior and neural activity that ismerely associated with the behavior. Therefore, it re-mains unclear whether the neural activation observedin previous studies reflects mechanisms that are vitalfor response inhibition (Garavan et al., 1999). In contrast,neuropsychological investigations can reveal whichneural regions are necessary for specific behaviors. How-ever, because such studies rely on patients with per-manent brain lesions, definitive conclusions regardingthe role of specific areas may be limited by the brain’scapacity to functionally reorganize following injury(Rorden & Karnath, 2004; Wall, Xu, & Wang, 2002).
The technique of transcranial magnetic stimulation(TMS) provides a unique opportunity to address theselimitations (Chambers, Payne, Stokes, & Mattingley,2004; Chambers, Stokes, & Mattingley, 2004; Robertson,Theoret, & Pascual-Leone, 2003; Walsh and Cowey, 2002).During TMS, a time-varying magnetic field is dis-charged over the scalp, causing temporary disruptionof underlying neural activity. As a reversible interferencetechnique, TMS can establish which cortical regions arevital for specific functions in the healthy brain, thuscomplimenting neuroimaging and neuropsychologicalmethods.
We used TMS to test the hypothesis that discrete regionsof the right hemisphere selectively govern responseinhibition in the healthy brain. Participants undertooka ‘‘stop-signal’’ task, which measured their ability toexecute and inhibit motor responses (Figure 1A; Logan,1994). On each trial, participants identified a target(‘‘go’’) stimulus as rapidly as possible (X or O) usingthe index finger of their left or right hand. On 25% oftrials, a ‘‘stop’’ signal was presented, instructing par-ticipants to withhold their response. To manipulate thedifficulty of successfully inhibiting, the stop signal waspresented randomly at various delays after the go signal.Previous studies have shown that the probability ofinhibition is closely related to this ‘‘stop-signal delay’’(SSD) (Figure 1B; Logan, 1981, 1994). Furthermore, be-cause this measure of inhibition is dependent on speedof responding, the SSD was adjusted according toeach participant’s mean reaction time (RT) (Figure 1C;Badcock et al., 2002).
Experiment 1 established psychophysical thresholdsof response inhibition for the left and right hands.Experiment 2 investigated the effects of deactivatingdiscrete regions of the right prefrontal and parietalcortex on inhibitory performance. Participants under-took the stop-signal task after receiving 15 min of TMS tothe right IFG (pars opercularis), MFG, or angular gyrus(AG) (Figure 2A; Table 1). TMS protocols of similarduration have been shown to suppress cortical excitabil-ity, thus temporarily deactivating the stimulated region(Siebner & Rothwell, 2003; Hilgetag, Theoret, & Pascual-Leone, 2001). To maximize the sensitivity of our TMSprotocol to changes in inhibitory performance, stopsignals were presented at SSDs that yielded 25–75%
correct inhibitions, as calculated in Experiment 1. Eachexperimental session involved stimulation of a differentanatomical region (IFG, MFG, AG), or a sham controlcondition. To measure any effects of cortical reorgani-zation over time, participants received two consecutiveblocks of TMS per session, each followed by the stop-signal task (Figure 2B, black shading). If inhibitory brainnetworks are able to compensate for the deactivation ofa primary region, then we expected cortical deactivationto be less effective following a second period of TMS.Finally, to determine whether a reduction of physiolog-ical arousal could explain any impairments of responseinhibition, pupil diameter was recorded throughoutthe experiment.
METHODS
Experiment 1: Stop-Signal Task
Seventeen right-handed volunteers were recruited (8 men,9 women, aged 18–27 years). Visual stimuli were pre-sented against a uniform gray background on a gamma-corrected Phillips Brilliance CRT monitor (19 in.; 1280 �1024 resolution; 100-Hz refresh rate). Each trial com-menced with the onset of a black fixation cross (0.68 �0.68; 100%). The visual target was a black ‘‘X’’ (2.68� 2.68)or ‘‘O’’ (2.68 diameter), presented at fixation (Figure 1A).A red box surrounding the target indicated a stop trial(3.98 � 3.98). White noise was delivered throughouttesting via two speakers positioned on either side of thevisual display. Participants also wore foam earplugs tomask ambient noise.
Inhibition functions were obtained over four to fivesessions of behavioral testing by use of an iterativemethod of constants. The first two sessions involvedSSDs of mean RT � 50, � 150, � 250, and � 350 msec.The remaining two to three sessions included SSDs atthe 25th, 45th, 55th, and 75th SSD percentiles, calculat-ed through sigmoidal regression of the results obtainedin Sessions 1 and 2. Psychophysical functions wereobtained with the three-parameter sigmoidal equation:
y ¼ a
1 þ e��
x�x0
b
�
Inhibition percentiles were obtained by solving forthe x values of 25, 45, 55, and 75 in the restructuredequation:
x ¼ � b � lna
y� 1
� �� x0
� �
Participants completed as many sessions as neces-sary to achieve a reliable inhibition function. The adjust-ed R2 for the sigmoidal regressions averaged .91 acrossparticipants.
Chambers et al. 445
Experiment 2: Stop-Signal Task followingTranscranial Magnetic Stimulation
Sixteen right-handed volunteers were recruited, all ofwhom had participated in Experiment 1 (8 men, 8 wom-
en, aged 18–27 years). To ensure the measurement ofthreshold-level inhibition performance, SSDs were pre-sented randomly at the 25th, 45th, 55th, and 75thpercentiles (mean RT � SSD) obtained from eachparticipant in Experiment 1. Participants wore foam
Figure 1. The stop-signal task
used to measure response
inhibition. (A) A typical display
sequence is shown for a‘‘stop’’ trial. On each trial,
participants identified a ‘‘go’’
signal (X or O) as rapidly aspossible using their left or right
hand. On 25% of trials, a
‘‘stop’’ signal (red box)
appeared around the targetfor 400 msec, signaling
participants to withhold their
response. The stop signal
could appear at various delaysfollowing the onset of the go
signal. In the example shown,
the participant correctlyinhibits, and the go signal
remains visible for 1000 msec.
On trials where participants
responded, the go and stopsignals disappeared and were
replaced by the intertrial
interval (gray cross). In all
experiments, the assignment oftarget (X or O) to hand (left or
right) was counterbalanced
between participants. (B) Thepredicted effect of the
stop-signal delay (SSD) on
inhibition performance. At
short SSDs, the stop signaloccurs soon after the onset of
the go signal, and participants
are able to inhibit easily (e.g.,
SSD of 50 msec; magenta-shaded area). As the SSD is
increased, participants are less
likely to successfully inhibit
because the go process iscloser to completion (e.g., SSD
of 250 msec; Logan, 1994). (C)
Inhibition performance in thestop-signal task depends on
the participant’s reaction time
(RT). To account for variation
in response speed, the SSDwas calculated with respect to
each participant’s mean RT,
and updated every 64 trials
within testing blocks. Thepattern of inhibition
performance yielded through
adjusted SSDs is the mirrorreverse of (B): As the SSD
approaches the participant’s
mean RT, the likelihood of
successful inhibition is reduced(green-shaded area).
446 Journal of Cognitive Neuroscience Volume 18, Number 3
earplugs and were delivered white noise throughoutthe experiment.
TMS/Magnetic Resonance Coregistration
Prior to Experiment 2, magnetic resonance (MR) brainscans were obtained from each participant using a GESigna 3T system (1.3 � 1.3 � 1.3 mm; sagittal acquisi-tion). To enable TMS/MR coregistration, participantswere scanned with contrast markers (vitamin E cap-sules) attached to known scalp locations (Chambers,Payne, et al., 2004; Chambers, Stokes, et al., 2004).Anatomical sites for TMS were then localized on thebasis of individual neuroanatomy. The IFG site wasdefined as the dorsal midpoint of the pars opercularis,between the lateral sulcus and inferior frontal sulcus(IFS), and directly anterior to the precentral sulcus. TheMFG site was defined as the dorsal midpoint of the MFG,between the IFS and superior frontal sulcus. The AG sitewas defined as the dorsal termination of the superiortemporal sulcus, which bifurcates the AG in the inferiorparietal lobule.
Average normalized coordinates for each site accord-ing to the Montreal Neurological Institute atlas areshown in Table 1. Scalp locations for TMS were calcu-lated using a magnetic tracking device (miniBird 500;Ascension Tech, Burlington, VT) and MR coregistrationsoftware (MRIReg).
TMS Parameters
TMS was delivered using a Magstim Rapid system (2.2 T,Magstim Company, Whitland, UK) and 70-mm figure-of-eight induction coil, fixed in position by a holding clampand tripod. The intensity of TMS was calibrated accord-ing to the maximum level of comfortable stimulation,expressed as a proportion of motor threshold, and ad-justed for differences in scalp–cortex distance betweenbrain regions (Stokes, Chambers, Gould, Henderson,Janko, Allen, & Mattingley, 2005). This protocol yieldedan average TMS output of 92% distance-adjusted motorthreshold. Consecutive testing sessions were separatedby at least 24 hr.
Figure 2. Magnetic stimulation sites and testing protocol in
Experiment 2. (A) Brain regions in the right hemisphere that werestimulated with TMS, shown for one participant. TMS was delivered
to the inferior frontal gyrus (pars opercularis; circle), middle frontal
gyrus (triangle) and angular gyrus of the parietal lobe (square).
Cortical sites were localized in each participant using sulcal landmarksfrom individual magnetic resonance (MR) scans. The location of the
TMS coil was then projected to the scalp using TMS/MR coregistration
(see Methods for details). Magenta line = precentral sulcus. Blueline = superior frontal sulcus. Orange line = inferior frontal sulcus.
Yellow line = lateral sulcus. Green line = superior temporal sulcus.
(B) The time course of each testing session in Experiment 2. Sessions
began with a practice block and pre-TMS block of trials, which wereused to obtain and update estimates of mean RT. Participants then
received 15 min of repetitive TMS followed by an experimental block
of 128 trials (post-TMS Block 1). After a short rest break, this TMS
protocol was repeated over the same anatomical site (post-TMSBlock 2). The order of TMS conditions (sham, IFG, MFG, AG)
between sessions was counterbalanced across 16 participants.
Table 1. Mean and Standard Deviation of NormalizedCoordinates (Millimeters) for Each Anatomical Location,According to the Montreal Neurological Institute Brain Atlas
BrainSite
Mean(x)
Mean(y)
Mean(z)
SD(x)
SD(y)
SD(z)
IFG 61 21 13 2.9 5.1 4.8
MFG 48 27 43 4.6 4.9 5.6
AG 53 �60 50 4.9 6.3 3.1
Chambers et al. 447
Sham Control Condition
The sham configuration provides a control condition inwhich the TMS coil is oriented away from the scalp,mimicking the sensory artifacts that accompany magnet-ic discharge without stimulating the cortex. Results inthe sham condition were collapsed across separateblocks in which the coil was placed over the parietalor prefrontal cortex. The order of sham placement(parietal, prefrontal) within sessions was counterbal-anced between participants.
Eye Tracking
Gaze was monitored online with an ASL-504 remoteinfrared eye tracker (ASL, Bedford, MA). Trials in whichparticipants blinked or gaze deviated more than 58 fromfixation were discarded. Pupil diameter in the right eyewas sampled every 20 msec (50 Hz) with a spatialresolution of 0.104 mm. Eye tracking ceased when theparticipant executed a response. On trials in whichparticipants responded (correct responses, failed inhibi-tions), analysis of pupil diameter was limited to the first900 msec of eye samples. Beyond 900 msec postfixationonset (300 msec posttarget onset), the variance of themean pupil diameter increased substantially across par-ticipants due to the increased likelihood of a response.
RESULTS
Experiment 1: Behavioral Patternsof Response Inhibition
Prior to administering TMS, we confirmed the validity ofthe stop-signal task by establishing psychophysical pat-terns of response inhibition in each participant. Asshown in Figure 3A, reducing the stop signal delayrelative to each participant’s mean RT improved inhibi-tion performance in both the left and right hands.Furthermore, as expected, participants responded sig-nificantly faster with their right hand (357 msec) thanwith their left hand (380 msec; p < .05). This executionadvantage for the right hand was mirrored by acorresponding inhibitory advantage. As indicated bythe drop lines in Figure 3A, participants required thestop signal to be presented significantly sooner in timerelative to their mean RT to successfully inhibit withtheir left hand compared with their right hand. Quanti-tative analysis of this effect is shown in Figure 3B. Pairedt tests revealed significant inhibitory advantages forthe right hand at SSDs that yielded 45% (mean advan-tage = 6 msec; p = .03), 50% (mean advantage = 7 msec;p = .01), 55% (mean advantage = 8 msec; p = .02), and75% correct inhibitions (mean advantage = 11 msec;p = .03).
In addition to demonstrating the accuracy of responseinhibition, Figure 3B indicates the latency of the in-hibition process, a term referred to as stop signal reac-
tion time (SSRT). The SSRT represents the theoreticallatency of inhibition by subtracting the SSD at whichparticipants correctly inhibited on 50% of trials fromtheir mean RT on ‘‘go’’ trials (mean RT � SSD50%;Badcock et al., 2002; Logan, 1994). The 50% point istheoretically important because it represents maximalcompetition between the go and stop processes. Asindicated by performance at the 50% percentile inFigure 3B, participants exhibited a significantly fasterSSRT with their right hand (201 msec) than with theirleft hand (208 msec).
Overall, the results of Experiment 1 confirm thevalidity of the stop signal task as a sensitive measure ofresponse inhibition. Furthermore, the presence of handdifferences in both inhibition and execution perform-ance underlines the importance of calibrating psycho-physical thresholds of response inhibition separately forthe left and right hands.
Experiment 2: Effects of Cortical Deactivationon Response Inhibition and Execution
In Experiment 2, participants completed the stop-signaltask following 15 min of repetitive TMS. Thresholdlevels of response inhibition were ensured by includingSSDs that yielded 25–75% correct inhibitions for eachhand in Experiment 1. To determine the effects of TMSon inhibition and execution performance, a variety ofbehavioral measures were analyzed. Execution abilitywas determined through analysis of mean RT andresponse accuracy on go trials. Inhibitory ability wasexamined by analyzing the latency and accuracy ofwithholding responses on stop trials. Because a suc-cessful inhibition has no observable latency, SSRT wasused to estimate the speed of the inhibition process(Logan, 1994). The accuracy of inhibition was deter-mined through analysis of the percentage of correctinhibitions.
Figure 4 reports inhibitory performance in the leftand right hands following sham TMS, or deactivation ofthe IFG, MFG, or AG. Figure 4A and C indicate resultsfollowing the first period of disruption (Block 1); b andd show the results following the second period ofdisruption (Block 2). A two-way ANOVA of inhibitionaccuracy in Block 1, with factors of TMS Condition andResponse Hand, revealed a significant main effect ofTMS Condition, F(3,45) = 3.6, p = .02 (Figure 4A).Analysis of simple main effects demonstrated a signifi-cant reduction in the percent of correct inhibitionsfollowing deactivation of the IFG relative to sham, inboth the left and right hands (both p < .05; stars inFigure 4A). No significant differences in inhibition accu-racy were observed between TMS Conditions of MFGand sham, AG and sham, or between IFG, MFG, and AG(all p > .28). Analysis of inhibition accuracy in Block 2revealed no significant effect of TMS Condition onbehavior, F(3,45) = 0.28, p = .84 (Figure 4B).
448 Journal of Cognitive Neuroscience Volume 18, Number 3
Figure 4C reports the average latency of responseinhibition in Block 1, as a function of TMS Conditionand Response Hand. A two-way ANOVA of meanSSRT revealed a robust main effect of TMS Condition,F(3,45) = 14.8, p < .00001. As indicated by stars inFigure 4C, this effect was driven by impaired inhibitoryperformance of the left and right hands following TMSof the IFG, compared with each of the sham, MFG,and AG conditions (all p < .015, Bonferroni corrected).Critically, deactivation of the MFG and AG yielded no
significant effects on inhibition performance relative tosham (all p > .8). Similarly, analysis of SSRT in Block 2revealed no significant main effect of TMS Condition,F(3,45) = 1.7, p = .17 (Figure 4D). These results dem-onstrate a significant effect of IFG deactivation on thelatency of response inhibition.
Figure 5 reports execution performance of the left andright hands following sham TMS, or deactivation of theIFG, MFG, or AG. As in Figure 4, left and right panelsillustrate the results following Block 1 or Block 2 of TMS,
Figure 3. Behavioral results
of the stop-signal task without
TMS (Experiment 1).
(A) Inhibitory performanceaveraged across 17 right-
handed participants.
Psychophysical inhibitionfunctions were calculated
using a three-parameter
sigmoid for the left (circles/
solid lines) and right (triangles/dotted lines) hands. As
expected, participants
inhibited more effectively at
shorter SSDs (larger values ofmean RT � SSD) than at longer
SSDs. Drop lines for left and
right hands denote the SSDsthat corresponded to 50%
inhibition performance. Note
that the function for the right
hand is shifted slightly in thenegative direction, indicating
improved inhibition. (B) The
SSDs required to yield
percentile levels of inhibitionin Experiment 1, plotted for
the left and right hands. Data
were calculated by fittingthree-parameter sigmoidal
regressions to each
participant’s inhibition
function and solving for the25th, 45th, 50th, 55th, and
75th percentiles. Error bars
are ±1 SEM.
Chambers et al. 449
respectively. Figure 5A and B present the mean RT onGo trials, whereas c and d show the mean percent ofassignment errors (responses with the wrong hand ongo trials). Separate two-way ANOVAs, with factors ofTMS Condition and Response Hand, revealed no signif-icant effects of TMS Condition on go RT or the rate ofassignment errors in either Block 1 or Block 2 (all F < 2.3,all p > .1).
Experiment 2: Relationship between InhibitoryDeficits and Arousal
As shown in Figure 4, analysis of the behavioral resultsrevealed a selective deficit of inhibition performance
following the first period of IFG deactivation. To whatextent might this observed deficit have arisen due toTMS-induced depression of arousal? (Karatekin, 2004;Niehaus, Guldin, & Meyer, 2001) To answer this ques-tion, we compared average pupil diameter in Block 1between TMS of the IFG and the sham control condi-tion (Figure 6). Crucially, no significant effects of TMSon pupil diameter were observed for trials in whichparticipants correctly inhibited (Figure 6A), failed toinhibit (Figure 6B), or responded correctly on go trials(Figure 6C). Furthermore, although pupil diameter in-creased as expected throughout the course of eachtrial (Karatekin, 2004), TMS of the IFG did not alterthe change in pupil diameter over time (Figure 6D–F ).
Figure 4. Inhibition performance in Experiment 2, averaged across 16 participants. (A, B) The percent of correct inhibitions following Block 1
(A) and Block 2 (B) of TMS, plotted by TMS Condition and Response Hand. As indicated by stars in (A), deactivation of the IFG significantlyreduced inhibitory performance relative to the sham condition, but only following Block 1. (C, D) Mean stop-signal reaction time (SSRT) following
Block 1 (C) and Block 2 (D) of TMS. Consistent with the accuracy results, deactivation of the IFG in Block 1 significantly slowed SSRT relative
to the sham condition. Error bars in all panels are ±1 SEM. Stars indicate a significance difference in performance between the respective TMScondition and the sham control ( p < .05).
450 Journal of Cognitive Neuroscience Volume 18, Number 3
These results indicate that the impairment of responseinhibition in Block 1 cannot be explained by diminishedarousal.
DISCUSSION
This study investigated the critical role of the rightprefrontal and parietal cortex in response inhibitionusing repetitive TMS. Results showed that temporarydeactivation of the IFG in the right hemisphere impairedinhibitory control of the left and right hands. Critically,TMS of the MFG and AG did not significantly alterinhibitory performance; nor did TMS of any regionssignificantly affect the speed or accuracy of responseson go trials. This dissociation of effects between stopand go trials enables us to eliminate various alternativeexplanations of the present results. It is possible, for
instance, that a deficit of response selection, responseexecution, or sustained attention could yield an inhibi-tory deficit (Sergeant, 2000). Note, however, that im-pairment of these functions would also be expected tocause a slowing of correct responses on go trials or anincrease in the rate of assignment errors (responses withthe wrong hand). Because neither result was observed,our findings suggest that deactivation of the IFG selec-tively impaired mechanisms responsible for inhibiting oroverriding prepotent responses.
Taken together, our results are consistent with theemerging view that executive control of response inhi-bition is mediated by ventral regions of the humanprefrontal cortex (Aron et al., 2004; Hasegawa et al.,2004; Morita et al., 2004; Aron, Fletcher, et al., 2003;Hazeltine, Bunge, Scanlon, & Gabrieli, 2003; Rubiaet al., 2003; Swainson et al., 2003; Durston, Thomas,
Figure 5. Execution performance in Experiment 2, averaged across 16 participants. (A, B) Mean RT on Go trials following Block 1 (A) and
Block 2 (B) of TMS, plotted as a function of TMS Condition and Response Hand. (C, D) The mean rate of assignment errors followingBlock 1 (C) and Block 2 (D) of TMS. No effects of TMS on response speed or accuracy were observed. Error bars in all panels are ±1 SEM.
Chambers et al. 451
Worden, Yang, & Casey, 2002; Sakagami et al., 2001;Garavan et al., 1999; Konishi et al., 1999; Kawashimaet al., 1996). Furthermore, the present study providesseveral unique insights into the cortical basis of inhibi-tory processing. First, because TMS selectively impairedresponse inhibition for both hands, our results indicatethat the right IFG fulfils an executive role in controllinginhibition in both cerebral hemispheres. Importantly,however, our results need not imply that the IFG is thesole mediator of response inhibition. The human IFG isrichly interconnected with a range of cortical and sub-cortical structures, including prefrontal regions in theopposite hemisphere, the anterior cingulate, and thestriatum (Vink et al., 2005; Durston et al., 2003). Conse-quently, deactivation of the IFG is likely to influenceprocessing in a variety of remote neural regions that alsocontribute to executive functions. Given the improbabil-ity of any executive function being mediated solely by asingle cortical subregion (Duncan & Owen, 2000), wefavor the view that the IFG is one critical componentwithin an inhibitory network. To further elucidate thedynamics of this network, future studies could combineTMS and neuroimaging to examine the effects of IFGdeactivation on neural activity in remote structures.
Second, our findings suggest that activation of theMFG and parietal cortex observed in previous neuro-imaging studies is unlikely to reflect processing that issingularly critical for response inhibition (Garavan et al.,1999; Kawashima et al., 1996). Instead, this activity mayreflect auxiliary processing within the prefrontal net-work. Alternatively, subsidiary activations revealed infMRI studies may reflect the execution of cognitiveprocesses that are activated in synchrony with responseinhibition but are not necessary for inhibitory control.This interpretation is consistent with a prominent reviewof neuroimaging studies by Duncan and Owen (2000).Through meta-analysis, these authors discovered thatmany prefrontal regions are activated during a variety ofexecutive functions, including response inhibition, re-sponse selection, and working memory. However, ratherthan indicating a generalized cognitive system, much ofthis coactivation may reflect the ecological likelihoodthat executive demands in one subsystem (e.g., workingmemory) are likely to require processing in another(e.g., response selection). Consequently, only a portionof these activations may be critical for mediating specificbehaviors, as suggested by the present findings andthose of Aron et al. (2004) and Aron, Fletcher, et al.(2003).
Third, our results showed that the effects of IFGdeactivation on response inhibition were specific forBlock 1 and did not arise in Block 2. This finding impliesthat although the IFG is critical for inhibitory processing,the cortical network that governs inhibition is able tofunctionally reorganize within approximately 30 minafter disruption, allowing critical processing withinthe IFG to be directed elsewhere (Siebner & Rothwell,2003). Based on previous studies, inhibitory mecha-nisms may be reallocated to the right MFG, parietalcortex, or homologous structures in the left hemisphere(Hester, Murphy, & Garavan, 2004). Hester et al. (2004)have shown that increasing the difficulty of responseinhibition yields additional activation of structures in theleft and right dorsolateral prefrontal cortex. Given thatTMS increased the difficulty of response inhibition in thepresent study, it is possible that these regions wererecruited in our participants to compensate for deacti-vation of the right IFG. Future studies could investigatethis question by varying the site of deactivation betweenblocks of TMS. If repetitive TMS can induce reorganiza-tion of executive processing, then ancillary brain regionsshould become vital for inhibitory control only followingdeactivation of a critical region.
Finally, our results show that neural mechanisms ofresponse inhibition and autonomic arousal can be effec-tively decoupled. Even though deficits of arousal, such asin ADHD and schizophrenia, are commonly accompa-nied by inhibitory pathology (Sergeant, 2005; Granholm& Verney, 2004; Hermens et al., 2004; Aron, Dowson,et al., 2003; Badcock et al., 2002), the present findingsindicate that impairments of inhibitory processing neednot be associated with deficits of autonomic arousal.This dissociation of TMS effects implies that the neuralsystems mediating inhibition and arousal are at leastpartially distinct.
The present findings open several avenues for furtherinvestigating the neural basis of response inhibition.For instance, given known behavioral interactions be-tween selection and inhibition of motor responses(Verbruggen, Liefooghe, & Vandierendonck, 2004), itwill be important for future studies to determine wheth-er specific prefrontal regions that are known to governresponse selection, such as the left dorsal premotorcortex (Praamstra, Kleine, & Schnitzler, 1999), are alsonecessary for inhibitory control. Furthermore, the in-creasingly feasible combination of simultaneous TMSand neuroimaging presents a unique opportunity toelucidate the architecture of critical and noncritical
Figure 6. The effect of IFG deactivation on arousal in Experiment 2. (A–C) Average pupil diameter following TMS of the IFG compared to the sham
condition. Data are plotted by target stimuli assigned to the left (white bars) and right (gray bars) hands. Results for the three most common
response types: correct inhibitions (A), failed inhibitions (B), and correct responses (C). (D–F ) Average change in pupil diameter throughoutthe course of each trial, after sham stimulation (circles) or deactivation of the IFG (squares). Results are collapsed across response hand and
plotted separately for correct inhibitions (D), failed inhibitions (E), and correct responses (F). The vertical dotted line in each panel indicates the
onset of the target stimulus (X or O). As indicated by the hatched areas in (E) and (F), results for failed inhibitions and correct responses are
truncated at 900 msec because eye tracking ceased when the participant responded. Error bars in all panels are ±1 SEM.
Chambers et al. 453
processing within the prefrontal cortex. A related objec-tive for TMS studies will be to disrupt multiple prefrontalregions simultaneously. Unlike single-coil stimulation,multicoil TMS can reveal whether brain areas that areunnecessary for inhibitory processing under normalcircumstances become vital during the simultaneousdeactivation of a primary region. Our results illustratethat the IFG of the right hemisphere is one such regionthat is crucial for inhibiting inappropriate action.
Acknowledgments
This work was supported by the Australian National Healthand Medical Research Council. We thank M. Rademacherfor technical assistance and A. Aron, J. Cooper, J. Duncan,R. Hester, C. Simoes, M. Williams, and two anonymousreviewers for helpful discussions.
Reprint requests should be sent to Christopher D. Chambers,Institute of Cognitive Neuroscience, University College Lon-don, 17 Queen Square, WC1N 3AR, UK, or via e-mail: [email protected].
REFERENCES
Aron, A. R., Dowson, J. H., Sahakian, B. J., & Robbins, T. W.(2003). Methylphenidate improves response inhibition inadults with attention-deficit/hyperactivity disorder.Biological Psychiatry, 54, 1465–1468.
Aron, A. R., Fletcher, P. C., Bullmore, E. T., Sahakian, B. J., &Robbins, T. W. (2003). Stop-signal inhibition disrupted bydamage to right inferior frontal gyrus in humans. NatureNeuroscience, 6, 115–116.
Aron, A. R., Robbins, T. W., & Poldrack, R. A. (2004). Inhibitionand the right inferior frontal cortex. Trends in CognitiveSciences, 8, 170–177.
Badcock, J. C., Michie, P. T., Johnson, L., & Combrinck, J.(2002). Acts of control in schizophrenia: Dissociatingthe components of inhibition. Psychological Medicine, 32,287–297.
Chambers, C. D., Payne, J. M., Stokes, M. G., & Mattingley, J. B.(2004). Fast and slow parietal pathways mediate spatialattention. Nature Neuroscience, 7, 217–218.
Chambers, C. D., Stokes, M. G., & Mattingley, J. B. (2004).Modality-specific control of strategic spatial attention inparietal cortex. Neuron, 44, 925–930.
Duncan, J., & Owen, A. M. (2000). Common regions of thehuman frontal lobe recruited by diverse cognitive demands.Trends in Neuroscience, 23, 475–483.
Durston, S., Thomas, K. M., Worden, M. S., Yang, Y., & Casey,B. J. (2002). The effect of preceding context on inhibition:An event-related fMRI study. Neuroimage, 16, 449–453.
Durston, S., Tottenham, N. T., Thomas, K. M., Davidson, M. C.,Eigsti, I.-M., Yang, Y., Ulug, A. M., & Casey, B. J. (2003).Differential patterns of striatal activation in young childrenwith and without ADHD. Biological Psychiatry, 53, 871–878.
Enright, S. J., & Beech, A. R. (1993). Reduced cognitiveinhibition in obsessive–compulsive disorder. British Journalof Clinical Psychiatry, 32, 67–74.
Garavan, H., Ross, T. J., & Stein, E. A. (1999). Right hemisphericdominance of inhibitory control: An event-related functionalMRI study. Proceedings of the National Academy ofSciences, U.S.A., 96, 8301–8306.
Goldman-Rakic, P. S. (1987). Circuitry of primate prefrontalcortex and regulation of behaviour by representational
memory. In F. Plum & V. Mountcastle (Eds.), Handbook ofphysiology: The nervous system (pp. 373–417). Bethesda:American Physiological Society.
Granholm, E., & Verney, S. P. (2004). Pupillary responses andattentional allocation problems on the backward maskingtask in schizophrenia. International Journal ofPsychophysiology, 52, 37–51.
Hasegawa, R. P., Peterson, B. W., & Goldberg, M. E. (2004).Prefrontal neurons coding suppression of specific saccades.Neuron, 43, 415–425.
Hazeltine, E., Bunge, S. A., Scanlon, M. D., & Gabrieli, J. D. E.(2003). Material-dependent and material independentselection processes in the frontal and parietal lobes: Anevent-related fMRI investigation of response competition.Neuropsychologia, 41, 1208–1217.
Hermens, D. F., Williams, L. M., Lazzaro, I., Whitmont, S.,Melkonian, D., & Gordon, E. (2004). Sex differences inadult ADHD: A double dissociation in brain activity andautonomic arousal. Biological Psychology, 66, 221–233.
Hester, R., Murphy, K., & Garavan, H. (2004). Beyondcommon resources: The cortical basis for resolving taskinterference. Neuroimage, 23, 202–212.
Heyder, K., Suchan, B., & Daum, I. (2004). Cortico-subcorticalcontributions to executive control. Acta Psychologica, 115,271–289.
Hilgetag, C. C., Theoret, H., & Pascual-Leone, A. (2001).Enhanced visual spatial attention ipsilateral to rTMS-induced‘‘virtual lesions’’ of human parietal cortex. NatureNeuroscience, 4, 953–957.
Karatekin, C. (2004). Development of attentional allocationin the dual task paradigm. International Journal ofPsychophysiology, 52, 7–21.
Kawashima, R., Satoh, K., Itoh, H., Ono, S., Furumoto, S.,Gotoh, R., Koyama, M., Yoshioka, S., Takahashi, T.,Takahashi, K., Yanagisawa, T., & Fukuda, H. (1996).Functional anatomy of GO/NO-GO discrimination andresponse selection—A PET study in man. Brain Research,728, 79–89.
Konishi, S., Nakajima, K., Uchida, I., Kikyo, H., Kameyama, M.,& Miyashita, Y. (1999). Common inhibitory mechanism inhuman inferior prefrontal cortex revealed by event-relatedfunctional MRI. Brain, 122, 981–991.
Logan, G. D. (1981). Attention, automaticity, and the ability tostop a speeded choice response. In J. Long & A. D. Baddeley(Eds.), Attention and performance IX. Hillsdale, NJ:Erlbaum.
Logan, G. D. (1994). On the ability to inhibit thought andaction: A users’ guide to the stop signal paradigm. SanDiego, CA: Academic Press.
Morita, M., Nakahara, K., & Hayashi, T. (2004). A rapidpresentation event-related functional magnetic resonanceimaging study of response inhibition in macaque monkeys.Neuroscience Letters, 356, 203–206.
Niehaus, L., Guldin, B., & Meyer, B. (2001). Influence oftranscranial magnetic stimulation on pupil size. Journalof the Neurological Sciences, 182, 123–128.
Praamstra, P., Kleine, B.-U., & Schnitzler, A. (1999). Magneticstimulation of the dorsal premotor cortex modulates theSimon effect. NeuroReport, 10, 3671–3674.
Robertson, I. R., Theoret, H., & Pascual-Leone, A. (2003).Studies in cognition: The problems solved and created bytranscranial magnetic stimulation. Journal of CognitiveNeuroscience, 15, 948–960.
Rorden, C., & Karnath, H. O. (2004). Using human brainlesions to infer function: A relic from the past era in thefMRI age? Nature Reviews Neuroscience, 5, 813–819.
Rowe, J. B., Toni, I., Josephs, O., Frackowiak, R. S. J., &Passingham, R. E. (2000). The prefrontal cortex: Response
454 Journal of Cognitive Neuroscience Volume 18, Number 3
selection or maintenance within working memory? Science,288, 1656–1660.
Rubia, K., Smith, A. B., Brammer, M. J., & Taylor, E. (2003).Right inferior prefrontal cortex mediates response inhibitionwhile mesial prefrontal cortex is responsible for errordetection. Neuroimage, 20, 351–358.
Sakagami, M., Tsutsui, K., Lauwereyns, J., Koizumi, M.,Kobayashi, S., & Hikosaka, O. (2001). A code for behavioralinhibition on the basis of color, but not motion, inventrolateral prefrontal cortex of macaque monkey. TheJournal of Neuroscience, 21, 4801–4808.
Sergeant, J. (2000). The cognitive-energetic model: Anempirical approach to attention-deficit hyperactivitydisorder. Neuroscience & Biobehavioral Reviews, 24,7–12.
Sergeant, J. A. (2005). Modeling attention-deficit/hyperactivitydisorder: A critical appraisal of the cognitive-energeticmodel. Biological Psychiatry, 57, 1248–1255.
Siebner, H. R., & Rothwell, J. C. (2003). Transcranialmagnetic stimulation: New insights into representationalcortical plasticity. Experimental Brain Research, 148,1–16.
Stokes, M. G., Chambers, C. D., Gould, I. C., Henderson, T. R.,Janko, N. E., Allen, N. B., & Mattingley, J. B. (2005). Simple
metric for scaling motor threshold based on scalp–cortexdistance: Application to studies using transcranial magneticstimulation. Journal of Neurophysiology, 94, 4520–4527.
Swainson, R., Cunnington, R., Jackson, G. M., Rorden, C.,Peters, A. M., Morris, P. G., & Jackson, S. R. (2003). Cognitivecontrol mechanisms revealed by ERP and fMRI: Evidencefrom repeated task-switching. Journal of CognitiveNeuroscience, 15, 785–799.
Verbruggen, F., Liefooghe, B., & Vandierendonck, A. (2004).The interaction between stop signal inhibition and distractorinterference in the flanker and Stroop task. ActaPsychologica, 116, 21–37.
Vink, M., Kahn, R. S., Raemaekers, M., van den Heuvel, M.,Boersma, M., & Ramsey, N. (2005). Function of striatumbeyond inhibition and execution of motor responses.Human Brain Mapping, 25, 336–344.
Wall, J. T., Xu, J., & Wang, X. (2002). Human brain plasticity: Anemerging view of the multiple substrates and mechanismsthat cause cortical changes and related sensory dysfunctionsafter injuries of sensory inputs from the body. BrainResearch Reviews, 39, 181–215.
Walsh, V., & Cowey, A. (2000). Transcranial magneticstimulation and cognitive neuroscience. Nature ReviewsNeuroscience, 1, 73–79.
Chambers et al. 455